Generating Books with LLMs

LLM-Generated Book about the Philosophy of AI

Introduction

This repository explores the application of Large Language Models (LLMs) to generate philosophical books, utilizing advanced AI models such as GPT-4 (gpt-4-0125-preview), OpenAI's GPT-3 (gpt-3.5-turbo-16k), and Google's Gemma (gemma:7b). The titles of the generated books are as follows:

• GPT-4: Mind, Machine, and Morality: Navigating the Philosophical Frontiers of Artificial Intelligence (PDF)

• GPT-3: The Ethical Nexus: Exploring AI's Philosophical Frontiers (PDF)

• Gemma-7b: The Turing's Legacy: Philosophical Intersections with Artificial Intelligence (PDF)

Additionally, I attempted to utilize the Anthropic Claude 3 Opus model. However, despite multiple attempts and experiences with their service, including encountering server errors and reaching token-per-day limits, obtaining a book from the Anthropic model proved unsuccessful. Despite these challenges, this repository showcases the capabilities and insights gained through the utilization of various LLMs in philosophical exploration.

Models Used

• OpenAI's GPT-3: Accessible via the OpenAI API.

• OpenAI's GPT-4: Accessed via the OpenAI API.

• Google Gemma: Accessed via OLLAMA (https://ollama.com/). To access the Gemma model, run the following command in your terminal:

  ollama pull gemma:2b

Description of Python Files

• app.py: The main file that orchestrates the entire application and generates the book.
• utils.py: Contains functions to instantiate the LLM models.
• structure.py: Generates a title, framework, and chapter list based on provided subject, genre, style, and profile of the book.
• ideas.py: Generates a list of ideas to be discussed in each chapter based on the generated title, framework, chapter list, subject, genre, style, and profile of the book.
• writing.py: Writes each chapter based on subject, profile, genre, style, chapter lists, and ideas. Sequentially writes about generated ideas, supporting arguments, and historical facts.
• publishing.py: Utilizes the Python docx library to generate the ".docx" file of the book.

Files Included

• .env: Contains information regarding API keys.

• requirements.txt: Lists the required libraries to be installed. To install them, run:

  pip install -r requirements.txt

Additional Resources

For more details on Langchain, Retrieval-Augmented Generation (RAG), and AI text generation consider enrolling in this Udemy course: Introduction to Langchain.